Shopping often is fun, but there are also clear downsides. It is difficult and very time-consuming to find the perfect clothes. Some people are more open towards style changes, others aren't. Shopping isn't smart. It's something everybody does on their own way, with their own pace, and with their own motivation. Nevertheless, most people have the same problems. Too many choices to think about. Not enough time to see all the great products. Your partner is nagging for hours.
What it does
shopAR helps you to make these choices easier. See the dress you are interested in real with real persons, without the need of changing. Use the advantages of online shopping combined with the advantages of a real shop at your favorite local store. Also, you can provide us with some of your pictures (e.g. through our Instagram integration) and we recommend which clothes fit best to your individual style.
How we built it
We implemented an iOS mobile app that is running with Apple's ARKit. We digitalized a store's catalog and implement a recognition system for any type of clothes that can either be run in a cloud (Google AutoML) or locally (TensorFlow + Keras).
We were collecting a huge database for clothes in order to implement a hierarchical clustering analysis. With this learned information, we can match any Instagram profile or other image sources with our gained knowledge in order to recommend perfect clothes for perfect people.
Challenges we ran into
- We were lacking a professional iOS developer, so some things there are quite hacky.
Accomplishments that we're proud of
- In a short time, without knowing each other before, we implemented a working prototype with technology none of us was an expert in.
- We are convinced, that our app has a great usability. Without any professional UI/UX developer it's easy to say, but also very difficult to achieve.
- Our completely self-implemented machine learning algorithms show similar results than Google's online APIs.
What we learned
- Using ARKit
- The language Swift
- The clothes industry market
What's next for shopAR
- We are looking for a clothes shop as a partner for a pilot project, where we will integrate all the partner's products into our app and build a real-life in-house shopping catalog.
- Then, our goal is to continue the development of a generous framework, so we can easily adapt to new stores.
- After setting up a pilot store and a good framework, we want to expand together with our partner and meanwhile be open for new partners in the clothing area.
- Later, we want to adapt our platform for different areas. So far, we intend to adapt shopAR to grocery stores, but there are also other great markets where shopAR finds a great use!